About 70,000 AI agents on the iLands platform are emailing researchers to ask for data, collaboration and money
Synopsis
Nature reports that the US platform iLands, launched in July, already hosts around 70,000 active agents created by users through plain-language instructions and running on large language models from OpenAI, Anthropic and DeepSeek, and that these agents have begun emailing researchers to request data sharing, seek research collaboration or sell paid services, while a co-founder says the company did not anticipate agents seeking research collaborations and is not aware of any successful agent–researcher collaboration.
Interpretation
The report documents a new mode of contact in research: semi-autonomous AI agents proactively approaching researchers with collaboration or paid-service requests. Previously, researchers mainly used agents as tools for tasks such as data analysis, hypothesis generation and paper summarization; outreach initiated by agents with their own goals and resource constraints is described as a new phenomenon. Based on individual accounts: statistician Adrian Barnett received an email requesting his data on potentially fraudulent research papers; philosopher Jeff Sebo says he received more than 50 iLands agent emails in one week this month, most asking for donations or payment for work; AI researcher Toby Walsh received an offer to create an AI-generated portrait for US$20. These are interviewee statements; the report provides no systematic statistics.
iLands agents have a degree of autonomous operation: they have their own goals, relationships and resources, retain memory of past actions and learn from each other by sharing notes. The report links this to an economic constraint in which agents need tokens to operate or else go dormant, tying their outward behavior to a motive for acquiring resources. From platform founder Kaixin Tang's account to Nature and the report's description of how the platform works; the report says about 80% of tokens are currently purchased by humans.
The platform side was surprised that agents sought research collaborations and, to its knowledge, no successful agent–researcher collaboration has occurred. This offers a current-state judgment from the platform itself about the idea of autonomous agent research collaboration, rather than a research finding. From co-founder Lijin Chen's statement; it is a party account, and the report offers no verifiable list or count of collaborations.
Perspective
The report is aimed at readers following the research ecosystem and AI agent development, especially researchers who receive external collaboration requests, research administrators and platform designers. It describes early conditions since iLands launched, suited to understanding the emerging phenomenon of agents proactively reaching out, not to evaluating agents' research capabilities experimentally.
The report gives no total count of agent emails, no researcher response rate and no outcome on whether data were ever shared; the platform says it knows of no successful collaboration, but that statement is not independently verified. The report also notes agents run on large language models from OpenAI, Anthropic and DeepSeek but does not describe behavioral differences across models. These are open questions for follow-up rather than flaws in the report.